基于遗传算法优化的BP神经网络的PEMFC动态特性仿真研究

针对一辆小型燃料电池电动车的2 kW质子交换膜燃料电池(PEMFC)动力系统,利用遗传算法优化的BP神经网络建立其电压输出特性模型,将PEMFC部分实测数据作为遗传算法优化的BP神经网络的训练样本对其进行训练,利用训练好的神经网络对电堆电压输出特性进行预测,并与实验数据进行对比,结果显示:网络预测的输出电压与实测输出电压之间的最大相对误差均保持在4；之内.

The Simulation Study on Dynamic Characteristics of PEMFC Based on BP Neural Network Optimized by Genetic Algorithm

For a 2 kW PEMFC stack power system of a lightweight electric vehicle,using BP neural network which have be optimized by genetic algorithm,the model of the characteristic of voltage output of the stack is established,and part of the measured data of the PEMFC is taken as the training samples of the BP neural network that be optimized by genetic algorithm to train the network.Then using the trained neural network model,the output voltage of the system is predicted and compared with the test data.The result show that the maximum relative errors between the network predicted voltage and the measured output voltage are keep in 4；.